Download as pdf or txt
Download as pdf or txt
You are on page 1of 25

Technology in Language Teaching & Learning

ISSN 2652-1687
https://www.castledown.com/journals/tltl/

Technology in Language Teaching & Learning, 6(1), 1–25 (2024)


https://doi.org/10.29140/tltl.v6n1.1142

ChatGPT as an AI L2 Teaching Support:


A Case Study of an EFL Teacher
MANUELA MENA OCTAVIOa

VICENTA GONZÁLEZ ARGÜELLOb

JOAN-TOMÀS PUJOLÀc

a b
The Language House Campo de Criptana, Spain Universitat de Barcelona, Spain
mmena30@gmail.com vicentagonzalez@ub.edu

c
Universitat de Barcelona, Spain
jtpujola@ub.edu

Abstract

The present study investigates the potential of ChatGPT as an L2 teaching support and aims to explore
the extent to which this Generative AI-powered tool is helpful for English as a Foreign Language
(EFL) teachers. Therefore, this exploratory single instrumental case study followed an EFL teacher
using ChatGPT for seven months, from January to June 2023, as an aid for her English classes in
a private language school in Spain. Data was collected from her ChatGPT history in relation to her
prompting tasks, from outputs received from her lesson plans when she implemented ChatGPT in
class, and from a semi-structured post-interview. The analysis followed a qualitative approach using
thematic analysis of the data to identify the type of support received by ChatGPT and to determine the
level of reliability and helpfulness of the outputs. The results indicate that ChatGPT offers significant
support to EFL teachers in planning and designing lessons, and implementing ChatGPT in lessons
and assessing learner’s writing. Our data highlights the potential of ChatGPT as a support for EFL
teachers as long as they possess specific AI competencies, such as prompt crafting to obtain useful
and meaningful outputs, and critical thinking, together with EFL teaching knowledge, to recognize
inaccuracies and identify relevant outcomes.

Keywords: Generative AI, ChatGPT, prompt crafting, critical AI literacy, EFL

Copyright: © 2024 Manuela Mena Octavio, Vicenta González Argüello & Joan-Tomàs Pujolà. This is an open
access article distributed under the terms of the C­ reative C
­ ommons Attribution Non-Commercial 4.0 International
License, which permits unrestricted use, d­ istribution, and reproduction in any medium, provided the original author
and source are credited. Data Availability Statement: All relevant data are within this paper.
ChatGPT as an AI L2 Teaching Support: A Case Study of an EFL Teacher 2

Introduction

With the release of ChatGPT in November 2022 Generative Artificial Intelligence (GenAI) has made
a significant impact on our society. However, this phenomenon is not entirely novel. As early as
2019, UNESCO published a comprehensive report highlighting the rapid expansion of AI and its
widespread implementation across various sectors of our lives. This report highlighted the potential
advantages of AI in education, including enhanced learning, assessments, and teaching support, while
also expressing concerns about its potential impact on the educational landscape, emphasising the need
for the regulation of AI development and implementation within the context of education (UNESCO,
2019).

In a later UNESCO document, Miao et al. (2021) define GenAI tools as artificial intelligence systems
designed to create new and original content from trained data. This can include generating text, images,
video, music, code and even 3D models. One such GenAI system is ChatGPT, which is large language
model (LLM) by OpenAI, a variant of the GPT (Generative Pre-trained Transformer) model, and is
trained on a massive amount of text data to generate human-like text. This chatbot has been trained
on a large dataset of text from the internet and can be used for a variety of tasks, such as generating
questions and answers, summarising text, translating from one language to another, and many others.
In fact, Sabzalieva and Valentini (2023, p. 9) summarise the multiple roles that ChatGPT could play
to augment teaching and learning, such as a collaboration coach, guide on the side, personal tutor,
co-designer, or motivator, among others.

ChatGPT in Education

Some authors have already shown that ChatGPT opens a wide field of possibilities for teaching
assistance and learning processes (Cooper, 2023; Gill & Kaur, 2023; Sabzalieva & Valentini, 2023;
A. Zhu et al., 2023). Most authors have concurrently warned us to be aware of some challenges
in education (Eke, 2023; Meyer et al., 2023). They state that all agents involved—technical
developers, policymakers in academic institutions, publishers, teachers, and students—must take
responsibility for ChatGPT’s good use and initiate a process of change to ensure a good integration
in the teaching context. It seems unavoidable that the advent of GenAI offers transformative
potential in the field of education. Although still in the early stages of an educational shift, we
have to assume that GenAI is already playing a crucial role in education (Cooper, 2023), and we
must draw up guidelines for using such GenAI tools and modify our teaching and assessment
practices accordingly (Kohnke et al., 2023).

Most of the studies conducted to date have also highlighted the potential drawbacks of the use of
ChatGPT, mainly due to ethical issues, information bias and low-quality information (Cooper, 2023;
Gill & Kaur, 2023; Meyer et al., 2023: Tenhundfeld & ChatGPT, 2023; Zhu et al., 2023). To date, some
outputs generated by ChatGPT suffered from a lack of accurate in-text citations, which serves as both
a limitation and a potential safeguard against the unrestricted use of AI to generate academic passages
(Tenhundfeld & ChatGPT, 2023). Román-Mendoza (2023) also points out the presence of inaccuracies
in some of ChatGPT’s answers in the context of Spanish as a foreign language education.

Apart from the aforementioned drawbacks, some studies provide information on the benefits that
ChatGPT can bring to education in general (Meyer et al., 2023) and to language education in particular
(Muñoz-Basols & Fuertes Gutiérrez, 2024; Peachey, 2023). After a SWOT analysis of ChatGPT,
Zhu et al. (2023, p. 149) propose a series of strategies that can harness the potential of ChatGPT in
3 Technology in Language Teaching & Learning, 6(2)

education: using ChatGPT a) as a virtual tutor for answering student questions; b) as an assistant for
creating learning materials; c) as expert support for completing complex tasks; d) as a virtual learning
buddy; e) as a reflective learning tool; and f) as a stimulus for critical thinking.

In the context of teaching Spanish as a foreign language, Román-Mendoza (2023) points out that
ChatGPT can be a good interlocutor for students to practise their language in interaction, but that the tool
still presents problems in some cases of lexical imprecision or even spelling. Román-Mendoza (2023)
adds that the strangest thing is that ChatGPT always seems to have to respond, even in some cases where
the student’s prompts are almost incomprehensible. All these problems may be because the development
of the tool in languages other than English is less developed at present (Rudolph et al., 2023).

ChatGPT offers great opportunities for teachers and educational institutions to improve L2 language
teaching. As stated by Hong (2023), we need to see ChatGPT as a learning tool and experiment with it
to explore its affordances. This tool can save teachers a lot of time. By automating teaching tasks such
as assessment and administration.

ChatGPT as an Assistant for Teacher Training

The good use of ChatGPT for teaching is necessarily linked to good teaching skills (Atlas, 2023;
Kohnke et al., 2023). Thus, Eke (2023) and Kohnke et al. (2023) point out the need to establish
training in the use of ChatGPT for both academic staff and students for optimal and responsible use.
Providing staff and students with the necessary support and resources can help mitigate potential risks
to academic integrity. Kohnke et al. (2023) identify specific digital competencies that teachers need to
develop to ensure an effective use of ChatGPT (Table 1).

Atlas (2023, p. 105) also provides valuable suggestions for educators aiming to leverage ChatGPT for
enhancing their professional growth: 1) experiment with the model in new and innovative ways; 2)
research the potential applications of ChatGPT in industries that have not yet been explored; 3) share
your insights and experiences with others through publications; 4) actively work to refine the model
and its capabilities by fine-tuning it; and consider mentoring others or creating educational resources
to help others learn about and work with ChatGPT.

Table 1 Specific forms of digital competence needed to use ChatGPT (Kohnke et al., 2023, p. 12)

Technological • Be aware of the features of ChatGPT


proficiency • Understand how ChatGPT works
• Construct effective prompts and interact with ChatGPT
• Troubleshoot challenges using ChatGPT in the classroom
• Stay up-to-date with changes with ChatGPT
Pedagogical • T
 hink about and plan ways to use ChatGPT to enhance or transform language
compatibility teaching and learning tasks
• Implement tasks that use ChatGPT
• Guide learners to use ChatGPT to use for self-directed learning
Social • H
 ave a critical awareness of the drawbacks of ChatGPT and consider them when
awareness planning and implementing tasks
• Inform learners of the risks, ethical issues, and drawbacks of ChatGPT
ChatGPT as an AI L2 Teaching Support: A Case Study of an EFL Teacher 4

The Current Study

The main aim of the present chapter is to investigate the implementation of ChatGPT as a Generative
AI teaching support tool by an English as a Foreign Language (EFL) teacher during a period of seven
months, from January to June 2023. The study seeks to explore the teacher‘s utilisation of ChatGPT to
enhance her language teaching tasks and her student’s language learning experiences in the classroom.
To achieve these general objectives, we formulated the following research questions:

• How did the EFL teacher integrate ChatGPT into her language teaching methods?
• How effective was ChatGPT in assisting the teacher in achieving her teaching goals?
• How did the teacher prompt ChatGPT to produce coherent and qualitative responses
relevant to her teaching goals?
• What strategies did the teacher employ to validate the reliability of the information
provided by ChatGPT?

It is hoped that the findings from this study will shed light on the potential benefits and limitations of
incorporating AI-powered tools like ChatGPT in the EFL classroom and provide valuable insights for
educators, researchers, and practitioners interested in AI-enhanced L2 language teaching.

Methodology

To address the research questions, this study adopts a qualitative approach (Dörnyei, 2007) of a single
instrumental case study (Stake, 2003) to gain insights into the use of ChatGPT by an individual EFL
teacher to support her teaching tasks. Thus, the research focus is on the potential of ChatGPT for EFL
teaching purposes.

Participants

The selection of the teacher participating in this study responds to the criteria of convenience or
opportunity (Dörnyei & Csizér, 2012). Researchers selected the teacher because she was one of the first
EFL teachers in Spain to use ChatGPT actively (free version, GPT3) for her teaching in January 2023.
Therefore, she agreed to share her chat history of ChatGPT and lesson plans for research purposes for
seven months and to be interviewed afterwards.

The teacher in the present study has over ten years of experience in teaching second languages, mainly
EFL and SFL, in various educational settings. Currently, she works at a private language school in
Spain where Spanish students of all ages and levels attend to prepare for official Cambridge ESOL
certifications. These students are the subsidiary participants of the study who are children (aged 4–12),
teenagers (aged 12–18) and adults (aged 19 and above). She teaches several small groups of different
English levels that range from A1 to C1 according to the Common European Framework of Reference
for Languages (Council of Europe, 2001) descriptors with an average of 8 students per group.

Data Collection

Data was collected from three sources: a) the teacher’s ChatGPT outputs in relation to her prompting
tasks, e.g. designing lesson plans, developing materials, designing activities or exercises, correcting
student’s writings, etc.; b) the lesson plans in which she implemented ChatGPT in her lessons; and
c) a semi-structured interview after the period of collecting data to analyse the teacher’s perception
of ChatGPT’s effectiveness in assisting her day-to-day teaching tasks and the strategies the teacher
5 Technology in Language Teaching & Learning, 6(2)

employed to validate the reliability of the information provided by ChatGPT. As content was the focus
of the interview, a verbatim transcription was made for ease of reading.

Data Analysis

The data gathered from the teacher’s ChatGPT interactions and lesson plans underwent a thorough
analysis utilising thematic analysis techniques and tasks from the teacher’s prompts. The data from
the interview was used to yield insights into the teacher’s perceptions of her usage of ChatGPT.
Additionally, the responses generated by ChatGPT were evaluated to assess their coherence, accuracy,
and relevance to the teacher’s prompts. The classification of this data into thematic categories followed
a bottom-up approach. Each researcher individually classified the typology of activities and prompts,
and then a discussion session took place to reach an agreement.

After the first analysis of the prompts the teacher used, the teacher’s actions and tasks using ChatGPT were
categorised and divided into three main groups: 1) lesson planning – the tasks that helped her to plan her
lessons; 2) lesson implementation – the tasks that she used ChatGPT with her students in the actual lessons;
and 3) assessment – the tasks related to evaluation processes, e.g. correcting student’s written assignments.

Results

Results are presented in three sections: focusing on the teacher’s prompt crafting strategies; the tasks
she performed using ChatGPT for her teaching purposes divided into the three categories mentioned
above (planning, implementation and evaluation), and finally, the strategies she used to check the
reliability of the outputs.
Prompt Crafting

Prompt crafting shows the ability of the teacher to ask ChatGPT to suit their teaching needs. Table 2
showcases the verbs used in the initial prompts crafted for lesson planning and assessment.

There is a clear evolution in the selection of verbs by the teacher for planning her classes, transitioning
from using the verb ‘suggest’ in January, when she was starting to interact with the tool, to employing
more specific action verbs such as ‘develop and provide’ or ‘prepare and encourage students to engage
in a lively debate’ in March, only two months later. Regarding the verbs for assessment, she started
using specific actions to create rubrics and provide feedback and ended up using ChatGPT only to
correct student’s compositions.

Apart from the specific actions, the teacher used ChatGPT three times to design comprehensive lesson
plans. These prompts were crafted as follows:

• Design a comprehensive lesson structure for an advanced language course to sequence


grammar, speaking, and writing exercises effectively. (January)
• Develop a lesson plan to work on Valentine’s Day in the English as a foreign language
classroom with intermediate-level students. Include activities and tasks to practice
all four language skills and take into account the 5Es: engagement, exploration,
explanation, elaboration, and evaluation. (February)
• Act as a language teacher preparing students for the FCE examination. Design a
60-minute lesson plan to help students prepare for the following speaking task: How
difficult do you think it is for people to be successful in these areas of life? Friendships
Work Parenting Sport Academic studies Talk together for about 2 minutes about the
questions and ideas provided in the task. (February)
ChatGPT as an AI L2 Teaching Support: A Case Study of an EFL Teacher 6

Table 2 Action verbs for the initial prompts

Month Specific Actions for Lesson planning (29) Actions for Assessment (12)
January • Suggest
• Design
• Propose
• Select
February • Choose and create • Create and include
• Present and provide • Analyse, evaluate and provide
• Choose and select feedback
• Develop and provide • Evaluate
• Create and include
March • P
 repare and encourage students to engage in a • Evaluate
lively debate • Generate
• Develop and craft follow-up questions • Create
• Choose • Develop
• Create (2)
• Present
• Explain
• Write (3)
• How can I teach
June • Integrate • Assess (5)
• Create (2)
• Develop
• Design (2)
• Provide (2)
• Generate

The specificity of prompts reveals that the teacher was trying to get more and better outputs. In Janu-
ary she asked for only one action to design a lesson plan but later in February prompted more actions
and asked for details to home in on her teaching goals and needs, considering the specifications of her
teaching context.

Likewise, the teacher’s general prompt crafting evolved from simpler outputs generated in the early
months of the study (January-February) to more complex ones, produced starting from March, when
the teacher began to implement a five-step strategy, based on her experience with ChatGPT and aimed
at fine-tuning her prompts, as depicted in figure 1 below.

• Fill in the blank with the appropriate preposition: “She’s interested learning
new languages.” (January)
• Present a scenario where students need to write an email or letter to a friend or
family member, asking for advice on a personal problem. Provide a writing task with a
word limit and a clear outline of the content and structure. (February)
• Act as my private language tutor and come up with the right answers for the following
gapped text (...) Provide a short explanation as to why those are the correct answers for
an FCE (B2) student of the language. (March)
7 Technology in Language Teaching & Learning, 6(2)

The task in the first example is limited to requesting an exercise such as a “fill in the blanks”; while
a month later, instructions were given for more complex tasks: creating a context (present scenario)
in which students had to activate a productive skill (write an email or letter) to a specific recipient
(friend or family) using specific communicative functions (asking for advice on a personal problem)
with limitations in terms of word count, content, and structure. This progression continued into the
following month of March when she asked ChatGPT to take on a specific role in designing the task for
the First Certificate of English type of students of a specific level (B2).

As can be seen in the examples above, experimenting with ChatGPT and observing the results in
the classroom allowed the teacher to make progress in the writing of her prompts, which led her
to discerning patterns in the student’s responses, identifying areas of challenge or proficiency, and
gaining insights into the function of the tool. All this knowledge helped her set up a five-step strategy
to craft her prompts (Figure 1).

As for the teacher’s use of ChatGPT in class with her students, a variety of activity types were requested,
as shown in Table 3. These prompts were created by the teacher but implemented by the students
with the intention of teaching them how to formulate prompts to utilise the AI tool as a personal
assistant in the second language learning process. The most frequent prompts used were those related
to Vocabulary (10), followed by Grammar (4), Writing (3), Reading (3) and Speaking (3).

There is a clear progression in the prompts they use in class. For instance, in relation to vocabulary, they
started by prompting only the specific meaning in a dictionary-like manner (the meaning of the idiom
‘spill the beans’). As time progressed, more complex actions were prompted, such as the creation of
vocabulary lists for more specific contexts, coherent with the student’s level, to include communicative
functions (greetings, ordering food, asking for directions, and making reservations). The same can be
said about prompts related to reading, writing, or oral production. By increasing the complexity of her
prompts in the classroom with her students, the teacher pursued a clear objective: to provide a variety
of different language strategies so that students can see the wealth of language learning affordances
ChatGPT offers.

In our data, Pronunciation and Listening Comprehension prompts were not used as ChatGPT 3.5 was
primarily designed for producing written texts, although there are browsers that can be used for oral
interaction with the tool that the teacher did not use.

Types of Tasks

The teacher prompted ChatGPT multiple tasks for her teaching purposes. Table 4 displays the v­ ariety
and frequency of tasks collected from January to June 2023 according to student’s age and level,
corresponding to the teacher’s classroom groups.

Figure 1 Enhancing Prompt Crafting: The Five-Step Strategy.


ChatGPT as an AI L2 Teaching Support: A Case Study of an EFL Teacher 8

Table 3 Student’s prompts for in-class implementation

Language Focus Prompt Samples


• F ill in the blank with the appropriate preposition: “She’s interested      
learning new languages.” (January)
• Correct the mistake in this sentence: “She don’t have enough time to finish the
assignment.” (January)
Grammar • What’s the correct form of the verb in this sentence: “If I       (have)
more time, I would have finished the project earlier.” (January)
• Act as my private language tutor and come up with the right answers for the
following gapped text. Provide a short explanation as to why those are the
correct answers for an FCE (B2) student of the language. (March)
• What’s the meaning of the idiom “spill the beans” (January)
• Use the word “ubiquitous” in a sentence to describe modern technology.
(January)
• Which phrasal verb can be used to complete this sentence: “I need to
Language the meeting due to a family emergency.” (January)
Systems • Compile vocabulary related to colours and basic shapes. (January)
• Create a list of words that describe various emotions and feelings. (January)
• Create a list of vocabulary related to daily routines, including activities that
can be done in the morning, afternoon, and evening (February)
Vocabulary
• Compile a vocabulary list of words related to travel, such as transportation
modes, accommodation types, and common phrases for asking directions.
(February)
• Create a vocabulary list of words and phrases that describe hobbies,
interests, and leisure activities. (March)
• Craft a list of useful phrases and expressions for travellers, such as greetings,
ordering food, asking for directions, and making reservations. (March)
• Create a list of adjectives and nouns that you can use to describe physical
appearance, personality traits, and emotions. Include some complex and
advanced sample sentences. Put all the information together in a table. (March)
• C an you explain the main idea of the first paragraph in this reading passage?
(January)
• What are the three reasons the author provides to support their argument in
Reading
this text? (January)
• According to this passage, how does the character feel about the situation
they are in? (January)
• D escribe your favourite book and explain why you enjoyed reading it.
(February)
• Write a formal email to your teacher explaining why you were absent from
Language
school yesterday. (February)
Skills Writing
• Create a story beginning with the sentence: “It was a dark and stormy night...”
(March)
• Compose a review of a movie or a book, giving your opinion on the plot,
characters, and overall experience. (March)
• R
 ole-play: Pretend we are at a restaurant, and I am a picky eater. Persuade
me to try a new dish from the menu. (February)
Speaking • Imagine we are planning a weekend trip. Suggest an itinerary and what
activities you would like to do. (March)
• Act as an X and have a chat with me about (topic). (March)
9 Technology in Language Teaching & Learning, 6(2)

Table 4 Frequency of tasks using ChatGPT

Tasks Children Teenagers Adults


(4–11) (12–18) (19+)
A1 A2 B1 B1 B2 C1 B1 B2 C1
Lesson planning
Curriculum and Content Suggestions – 2 – – – 1 – – –
Activity Generation – 4 3 2 3 2 2 3 1
Lesson Structure and Sequencing – – – 1 – – 1 – –
Cultural Insights – – – 1 1 – – – –
Vocabulary Expansion 2 1 – – – – – – 2
In-Class Lesson Implementation
Real-time Q&A – – – 3 1 – – – –
Interactive Language Practice – – 1 2 4 2 – – –
Sentence and Text Editing – – 1 2 1 – – – –
Vocabulary lists & tables 1 2 1 1 – 1 – 1 –
Assessment
Automated grading, evaluation and feedback – 1 1 1 1 1 1 1 1
Language proficiency tests – 2 – – – – – – –
Reading comprehension tasks – 1 – – – – – – –
Creative assessment – 1 – – – – – – –

There were a total of 68 productions generated by ChatGPT, divided as follows:

• 
24 prompts aimed at teaching children, 12 of which were devised for planning, 6 for
in-class lesson implementation, and 6 for evaluation.
• 
31 prompts aimed at teaching teenagers, with 11 for planning, 17 for in-class lesson
implementation, and 3 for evaluation.
• 
13 prompts aimed at teaching adults, including 9 for planning, 1 for in-class lesson
implementation, and 3 for assessment.

The differences in the frequency of ChatGPT usage based on age and level are primarily due to the
teacher’s conclusion that the AI tool could be valuable for generating lesson plans, texts, materials,
and other resources of low complexity, adapted to the age and proficiency level of learners at the ele-
mentary or intermediate level. In her experience, the texts and materials produced for advanced levels
lacked relevance, deviated from the intended objectives, or were too superficial.

Furthermore, as the teacher stated in the interview, the materials contained in textbooks for children
are often very limited. The use of ChatGPT represented an opportunity to expand and enhance didactic
resources for classroom use. According to her experience, ChatGPT provides creative and easy
solutions for children in the classroom, something that is not commonly found in textbooks.
ChatGPT as an AI L2 Teaching Support: A Case Study of an EFL Teacher 10

Regarding the differences in frequency in the typology of actions, the number of prompts proposed for
assessment is considerably lower, as minor modifications to the same prompt can be used to correct
compositions at any level. Additionally, continuous assessment is a highly specific procedure that
requires taking numerous variables into account. On the other hand, the increase in the number of
prompts for classroom implementation is due to two main reasons. Firstly, the teacher observed a high
willingness on the part of adolescent students to use AI in the classroom. Secondly, the teacher aimed
to demonstrate to her students how to independently work with the L2 in context, using prompts that
were written and verified by her.

Lesson Planning

The Lesson Planning section showcases how ChatGPT has become an indispensable ally for L2 teach-
ers, as it considerably alleviates their workload. In the case of the current study, the EFL teacher asked
the AI tool to perform the following tasks and actions to help her with her planning and designing
activities:

1. Curriculum and Content Suggestions – ChatGPT provided ideas for:


• lesson topics,
• relevant materials, and
• suggested activities to align with the curriculum.
2. Activity Generation – ChatGPT generated:
• grammar and vocabulary exercises
• quizzes and games
• worksheets to reinforce language concepts or grammar rules.
3. Lesson Structure and Sequencing – It helped in:
• organising the flow of the lesson and
• suggested how to sequence different activities for better learning outcomes.
4. Cultural Insights:
• It provided information about the culture and customs related to the language being
taught.
• It enabled the teacher to create culturally enriched lessons.
5. Vocabulary Extension:
• It suggested new vocabulary words and phrases suitable for various proficiency
levels.
• It aided in broadening student’s lexical repertoire, tailoring new word suggestions to
suit various proficiency levels in the blink of an eye.

Below are two illustrative instances of tasks created by ChatGPT for lesson planning. These selected
examples effectively highlight the tool’s capacity to bridge the gap between intricate definitions and
the student’s cognitive and linguistic capabilities, as evidenced by the first case (Excerpt 1). In addi-
tion, as can be seen in the prompt provided to the ChatGPT in Excerpt 1, in March, four key steps of
the 5-step strategy were already incorporated to achieve more effective outputs.

Excerpt 1.

Example of Lesson Planning (lesson structure). Initial prompt: Can you explain the following con-
cepts to my seven-year-old students? Matter, States of Matter and Properties of Matter.
11 Technology in Language Teaching & Learning, 6(2)

The output offered by ChatGPT is in line with the teacher’s request and the age and level of the stu-
dents. This is why the teacher stated in the interview that:

ChatGPT played a crucial role in helping me brainstorm ideas and approaches for my
lesson plans, especially for kids. Its ability to generate diverse content and examples
allowed me to create lessons that were not only relevant to my student’s interests but
also catered to their individual learning needs. Additionally, ChatGPT’s suggestions
helped me add variety and creativity to my teaching materials, making the learning
experience more enjoyable and productive for my students.

Excerpt 2 below shows the prompt elaborated by the teacher in June, being specific and fulfilling four
steps, which indicates that she reached a level of mastery in the use of ChatGPT, and thus, the expected
quality of the output. Equally noteworthy in this excerpt is the tool’s skill in presenting information
through visual synthesis and employing contextually fitting examples, such as songs or raps, to facili-
tate vocabulary acquisition in a manner that resonates with the core tenets of language instruction for
young learners as exemplified.
ChatGPT as an AI L2 Teaching Support: A Case Study of an EFL Teacher 12

Excerpt 2.

Example of Lesson Planning (Vocabulary extension). Initial prompt: Generate a thematic vocab-
ulary list for young learners, focusing on daily routines and activities. Include meaningful examples
and a rap song.
13 Technology in Language Teaching & Learning, 6(2)
ChatGPT as an AI L2 Teaching Support: A Case Study of an EFL Teacher 14

In addition, in the case of Excerpt 2, we can also see the capacity for creation, originality and coherence
that the output provides. The examples of use offered are coherent throughout the actions requested by
the prompt: the thematic vocabulary list offered is the one included in the examples, and the same one
is integrated in the song.

Both creations were introduced into the classroom and yielded impressive results, as shared by the
teacher during the interview:

ChatGPT’s assistance in developing instructional materials was invaluable. It provided


me with a vast range of resources and authentic language examples that I could use
to address different language proficiency levels among my students. With ChatGPT’s
support, I could tailor materials to suit both struggling learners and those who needed
additional challenges. This personalization improved student’s engagement and
facilitated their language acquisition process.

The resources proved to be immensely helpful in enhancing the student’s comprehension and retention
of new vocabulary and concepts.

In-Class Lesson Implementation

ChatGPT in-class lesson implementation empowered the EFL teacher by enabling real-time Q&A and
interactive language practice. With its sentence and text editing capability and its vocabulary lists and
tables feature, ChatGPT supported her students in widening their vocabulary as well as honing their
writing skills, enhancing sentence structure, and refining compositions.

1. Real-time Q&A – ChatGPT assisted in:


• answering student questions during the lesson,
• providing immediate clarifications and explanations.
2. Interactive Language Practice – It engaged in:
• simulated conversations,
• role-playing, and
• language practice with students to improve their speaking/writing (productive)skills.
3. Sentence and Text Editing – ChatGPT helped students by suggesting improvements to
their sentences and compositions.
4. Vocabulary lists & tables – It generated:
• comprehensive and tailored vocabulary lists, categorising words based on themes,
topics, or difficulty levels,
• organised tables that display words, their meanings, example sentences, and relevant
contextual information.
5. Reading comprehension – ChatGPT:
• generated a text requiring students to comprehend and condense the main ideas,
• asked students to identify the primary concepts or themes in a passage, helping them
grasp the central message,
• had students infer the meaning of specific words or phrases from the context of the
passage,
• challenged students to predict what might happen next in the story based on the
current events in the passage,
• generated sentences with missing words and asked students to fill in the blanks with
appropriate words based on the context,
• had students relate the themes or concepts from the passage to real-world situations
or their own experiences.
15 Technology in Language Teaching & Learning, 6(2)

Below are two examples of outputs generated by ChatGPT based on two prompts crafted by the teacher
during the study (Excerpt 3 and Excerpt 4). The examples also demonstrate an evolution in terms of
prompt formulation, as can be observed in both excerpts:

Excerpt 3.

Example of In-Class Implementation (Vocabulary Extension). Initial prompt: Compile vocabulary


related to colours and basic shapes.
ChatGPT as an AI L2 Teaching Support: A Case Study of an EFL Teacher 16

ChatGPT not only provided the requested information but also gave the teacher advice on how to bring
the vocabulary to the classroom and facilitate both understanding and learning (similar suggestions
that also appeared in Excerpts 1 and 2), which the teacher also outlined in the interview:

By incorporating ChatGPTs’ suggestions and advice, I noticed an increase in student


participation and enthusiasm for language learning.

In Excerpt 4, we observe that as the teacher became more familiar with the tool, she increased the
complexity of what she asked for in terms of formal aspects of the language.

In this case, she asked ChatGPT to give her the information in a spreadsheet so that she could take it
back to the classroom as it was.

Excerpt 4.

Example of Lesson Planning (Vocabulary Extension). Initial prompt: Create a list of adjectives and
nouns that you can use to describe physical appearance, personality traits, and emotions. Include
some complex and advanced sample sentences. Put all the information together in a table.

The progression in prompt crafting is notably discernible. As exemplified in Excerpt 3, when guided
by the provided instructions, the tool generates a mere list of terms, which offers limited value in terms
of acquiring new vocabulary in a second language. Conversely, Excerpt 4 unveils a meticulously
organised table categorised by usage examples.
17 Technology in Language Teaching & Learning, 6(2)

This method proved instrumental in facilitating substantial and enduring comprehension among students,
a conclusion reaffirmed through subsequent objective assessment tests administered by the teacher.

Assessment

The assessment section highlights ChatGPT’s pivotal role in objective grading through automated
grading, providing valuable writing evaluation feedback, and assessing student’s written production
and reading comprehension skills. Moreover, its ability to generate language proficiency tests, tailored
to specific language levels, enhanced the comprehensive evaluation of student’s language abilities. By
utilising these ChatGPT features, the teacher could streamline assessment processes and offer more
tailored feedback to students. The teacher used ChatGPT for the following tasks on assessment in two
ways:

Category 1:

Automated Grading, Evaluation, and Feedback (Tasks 1, 2 and 3)

These tasks involved the automated assessment and evaluation of student’s work, along with provid-
ing feedback. Both tasks centred around leveraging ChatGPT’s capabilities to streamline the grading
process and enhance the efficiency of evaluating student’s language skills.

1. Automated grading, evaluation, and feedback – ChatGPT was used to:


• automatically grade assignments, essays, or other written tasks submitted by
students; create prompts;
• design assessment rubrics;
• assess the material, grammar, word choice, and flow by devising imaginative and
captivating evaluation exercises. For instance, educators could prompt students to
craft a dialogue between two figures from history, fashion a poem, or create a brief
narrative employing targeted vocabulary and grammatical constructs; and
• provide instant feedback, highlighting areas that need improvement.
2. Language proficiency tests – ChatGPT:
• generated language proficiency tests for students at different levels. These tests can
cover various language skills, such as listening, speaking, reading, and writing; and
• created questions that assess grammar, vocabulary, comprehension, and more;
• gauged student’s progress, identified areas of improvement, and tracked their
language development over time.
3. Creative assessment – ChatGPT assisted in:
• providing prompts that encourage students to showcase their language skills in
imaginative ways, making the assessment process more enjoyable and fostering
creativity in language learning.

To ascertain the credibility of ChatGPT in the assessment and scoring of writings created by students
for the Cambridge assessment, a strategic approach was adopted. This involved employing the correc-
tion models outlined in the official Cambridge Handbook for Teachers (2023).

By doing so, the outcomes generated were systematically juxtaposed with the revisions carried out
by adept examiners who possess proper training and accreditation. These examiner corrections are
outlined by Cambridge University Press & Assessment (2023, p. 39). It should be noted that the prompt
is much more specific than the ones so far and that it is in compliance with the 5-step strategy.
ChatGPT as an AI L2 Teaching Support: A Case Study of an EFL Teacher 18

Excerpt 5.

Example of Assessment. Initial prompt: This is an essay written by an adult FCE (B2) candidate
based on the following task: In your English class you have been talking about the environment (...)
Assess the essay based on the Cambridge FCE writing scales and give a mark out of 5 for each of the
evaluated items (Communicative achievement, Language, Content, Organisation & Overall), plus a
general mark out of 20 (...)
19 Technology in Language Teaching & Learning, 6(2)

The output of Excerpt 5 shows a discrepancy of three points between the official correction and the
assessment offered by ChatGPT in the final score. Specifically, in the ‘Content’ category, the tool
attributed a score of 3, diverging from the official score of 4. Similarly, within the ‘Language’ segment,
the tool conferred a score of 2 in contrast to the officially designated 3, as indicated in the official
handbook. Furthermore, within the ‘Organization’ aspect, ChatGPT assigned a score of 2 instead of
the established official score of 3 to the candidate.

Category 2:

Student Self-Assessment (Task 4)

These kinds of tasks were for students to evaluate their own learning in more creative aspects of lan-
guage, like understanding what they read, thinking critically, and expressing themselves creatively.
These tasks focused on helping students understand and use language more deeply.

4. Self-assessment on reading comprehension and writing – ChatGPT:


• designed a portfolio to help students self-assess on reading comprehension and writing;
• created multiple-choice questions that assess comprehension of details, characters,
or events in the reading;
• generated statements based on the passage and asked students to determine if they
were true or false;
• crafted questions that required students to use new vocabulary words correctly in
sentences related to the passage;
• had students arrange sentences or events from the passage in chronological order;
• created questions that prompt students to analyse the personalities, motivations, and
actions of characters in the reading;
• created questions that required students to interpret metaphors, symbolism, or
figurative language in the passage; and crafted questions that prompted students to
evaluate the author’s arguments, perspectives, or intentions.
ChatGPT as an AI L2 Teaching Support: A Case Study of an EFL Teacher 20

Excerpt 6 below is an illustrative instance of a creative assessment output ingeniously generated by


ChatGPT. As depicted, when the goal is to forge creative assessment tools tailored to a specific sub-
ject matter, the tool emerges as an invaluable ally. Both the complexity and substance of the portfolio
tasks harmoniously resonated with the student’s language proficiency level and the predetermined
objectives. These objectives, which had been originally crafted by ChatGPT at the beginning of the
instructional unit, underscored the tool’s foresight in meeting educational objectives.

Excerpt 6.

Example of Assessment. Initial prompt: Can you create a portfolio to help my students (aged 6–7 with
a beginner level of the L2) self-assess their knowledge and understanding of the subject of states and
properties of matter?
21 Technology in Language Teaching & Learning, 6(2)

The use of ChatGPT in assessment tasks for students satisfied the teacher due to the creative alterna-
tives it offered, as seen in the portfolio proposal presented in Excerpt 6. The sections presented, the
incorporation of emoticons, and the use of clear and appropriate language for the age and level of
the students should be highlighted. The teacher pointed out the following in the interview:

I used ChatGPT outputs as a reference to provide targeted feedback to my students.


While I always ensured to review and validate the suggestions, ChatGPT’s assistance
saved me time and allowed me to focus more on providing personalised feedback to
each student.

Reliability of Outputs

The quality of the information offered by ChatGPT, its veracity, and the possible biases in terms
of gender, origin or any other type should also be considered. Developing critical thinking about
the outputs provided is a strategy that the teacher considered from the start. In the interview, she stated
the strategies she employed to validate the reliability of the information provided by ChatGPT.

Firstly, we encouraged students to cross-reference the information provided by ChatGPT


with reputable and credible sources. By comparing the information from multiple
reliable sources, we could identify any discrepancies or inaccuracies.
ChatGPT as an AI L2 Teaching Support: A Case Study of an EFL Teacher 22

It is noteworthy that she incorporated her students in the tasks of reviewing the information.

Secondly, we fact-checked the information generated by ChatGPT using reputable


fact-checking websites or databases. This helped us verify the accuracy of the data and
claims made by ChatGPT.

She also involved other colleagues from her school in the task of reviewing the information.

In addition, teachers, as well as students involved in the experiment, were encouraged to


review each other’s work and provide feedback. This collaborative approach allowed for
critical evaluation of the information generated by ChatGPT from different perspectives.

This review considered possible biases as well as the veracity and quality of the information.

We were aware that AI language models like ChatGPT may have inherent biases in
their training data. To address this, we actively looked for any biases in the generated
content and corrected or supplemented it with a more balanced perspective when
needed.
I supervised the output of ChatGPT closely to ensure that the language used was
appropriate, coherent, and grammatically correct. Moreover, the questions or prompts
given to ChatGPT were carefully crafted to elicit accurate and relevant responses.
Ensuring that the prompts were based on accurate information contributed to the
overall reliability of the responses.

This process of reviewing the information, either alone, with her students or with other colleagues
is useful as a guideline for teachers who want to introduce the use of ChatGPT but fear possible
disinformation risks.

Discussion

The data analysis carried out shows us the potential ChatGPT has for teaching assistance and learning
processes, as stated by several authors (Cooper, 2023; Gill & Kaur, 2023; Sabzalieva & Valentini, 2023;
Zhu et al., 2023). However, as the teacher admitted during the interview, her journey in learning to bet-
ter use ChatGPT prompts was a process of exploration and experimentation. Initially, she familiarised
herself with the capabilities and limitations of ChatGPT, understanding its underlying technology and
purpose in the L2 generation. She spent some time exploring ChatGPT through various platforms and
interfaces, engaging in practice sessions to observe how it responded to different prompts, both generic
and specific to language learning topics. Carefully reviewing sample responses generated by ChatGPT
allowed her to analyse the accuracy, clarity, and relevance of the answers, enabling her to gauge its
performance and identify areas where she needed to guide her students more effectively.

The teacher’s approach aligns with Hong’s (2023) ideas about exploring ChatGPT’s potential as a
learning tool and teacher’s assistant. Through persistent experimentation, the teacher identified effec-
tive initial prompt steps. This experimentation highlighted ChatGPT’s potential for various teacher
tasks like lesson planning, in-class teaching, and assessment. Utilising ChatGPT for tasks such as auto-
mated grading and generating language tests can save time and ensure consistent evaluation, aiming to
objectively assess student language proficiency and assist in self-assessment.

The teacher was also careful in verifying the quality and reliability of the information generated by
ChatGPT, and she was in need to check that there was no bias of any kind, as proposed by UNESCO
23 Technology in Language Teaching & Learning, 6(2)

(2019), Tenhundfeld & ChatGPT (2023), or Zhu et al., (2023), among others. Moreover, these checks
were formative in nature (Atlas, 2023) for both the students and the other colleagues in the Language
Centre. The teacher specifically tried to engage students with the aim of getting them used to reading
the outcomes critically.

In addition to using ChatGPT to prepare lesson plans and materials, the teacher also implemented the
AI tool during sessions to empower the students and grant them greater autonomy. This demonstrated
to the students the possibilities GenAI offers for self-study and the development of communicative
competence autonomously.

It seems necessary for a good use of ChatGPT for L2 teaching that teachers develop their own strat-
egies for its use to meet their teaching objectives (Atlas, 2023; Kohnke et al., 2023). Following the
specific forms of digital competence underlined by Kohnke et al. (2023) – technological proficiency,
pedagogical adaptability, social awareness, critical awareness – the teacher adeptly assimilated the
features of ChatGPT. She demonstrated a comprehensive understanding of its mechanisms and exhib-
ited a strong proficiency in prompt crafting (technological proficiency). In terms of pedagogical adapt-
ability, she displayed the ability to strategize varied didactic uses of ChatGPT, particularly in lesson
structuring, classroom integration, and assessment methodologies. Moreover, she proposed methods
for her students to independently utilise ChatGPT. Consistently throughout the process, she demon-
strated a strong sense of social awareness concerning the content generated by ChatGPT. This was
exemplified by her critical awareness and engagement with the tool. Her mindful approach not only
acknowledged the technology but also considered the broader societal impact and ethical dimensions
associated with the use of ChatGPT in an educational context.

Conclusion

Following one EFL teacher for seven months allowed researchers to observe the evolution of her
strategies using ChatGPT as they improved over time and, thus, the effectiveness of ChatGPT for her
teaching purposes. She demonstrated a skillful ability to use the tool and develop most of the forms
of digital competence needed to use ChatGPT according to Kohnke et al. (2023) (Table 1). Moreover,
two fundamental strategies were clearly identified that every teacher and student should develop when
using ChatGPT: prompt crafting and critical thinking.

After analysing the results and regarding our first question, we can affirm that the teacher managed
to integrate ChatGPT in the teaching processes carried out in her classes. This was achieved by
experimenting with the writing of prompts. These prompts started out simple (no more than one
action per prompt), especially in the request for lesson planning or tasks in the month of January until
reaching an optimal initial prompt structure throughout the months. This structure was based on the
incorporation of 5 steps for useful information: action, format, context, target, and extra information. It
is the wording of these prompts that enabled ChatGPT to generate coherent and qualitative responses
relevant to the teacher’s goals, as seen in the six excerpts. Moreover, thanks to the information provided
by the teacher throughout the interview, we know that ChatGPT was effective in helping the teacher
design most teaching tasks for her classes of different ages and levels. In some excerpts, we were able
to appreciate how ChatGPT offered original ideas to facilitate the management or implementation of
those tasks in the classroom. Regarding precautions to be considered before the information generated
by ChatGPT, we have to say that the teacher deployed a range of useful strategies that helped to
ensure the reliability of the information provided by ChatGPT. These strategies focused on the contrast
and verification of the information provided by ChatGPT by all the agents involved in the teaching-
learning process: the students, the teacher herself and the rest of the teachers of the Language Center
where she works.
ChatGPT as an AI L2 Teaching Support: A Case Study of an EFL Teacher 24

The inferring limitations of a single case study cannot offer anything beyond this particular case.
However, the relevant information achieved in this exploratory empirically-rich study of a selected
active teacher using ChatGPT opens the door to conducting new larger-scale studies that take the
results of the present study into account.

ChatGPT emerges as an invaluable resource, revolutionising L2 classrooms by elevating three fun-


damental dimensions: lesson planning, in-class lesson implementation, and assessment. The GenAI
language model’s multifaceted capabilities can empower language educators, providing them with
dynamic tools to create engaging lessons, deliver interactive teaching experiences, and conduct com-
prehensive evaluations. In exploring the transformative role of ChatGPT, we unveil its potential to
shape more efficient and enriching L2 learning journeys for students and teachers alike. Teachers need
to experience its potential for teaching, to share the results with other colleagues, and to contrast them
with recently published research, as our teacher conscientiously did.

References

Atlas, S. (2023). ChatGPT for higher education and professional development: A guide to conversa-
tional AI. Amazon Digital Services LLC. https://digitalcommons.uri.edu/cba_facpubs/548
Cambridge University Press & Assessment (2023). B2 First Handbook. https://www.cambridgeenglish.
org/images/167791-b2-first-handbook.pdf
Cooper, G. (2023). Examining science education in ChatGPT: An exploratory study of generative
artificial intelligence. Journal Science Education and Technology, 32, 444–452. https://doi.
org/10.1007/s10956-023-10039-y
Council of Europe (2001). Common European Framework of Reference for Languages: Learning,
Teaching, Assessment. Cambridge University Press.
Dörnyei, Z. (2007). Research methods in applied linguistics: Quantitative, qualitative and mixed
methodologies. Oxford University Press.
Dörnyei, Z. & Csizér, K. (2011). How to design and analyze surveys in second language acquisition
research. In A. Mackey & S. M. Gass (Eds.), Research methods in second language acquisition:
A practical guide (pp. 74–94). Wiley. https://doi.org/10.1002/9781444347340.ch5
Eke, D. O. (2023). ChatGPT and the rise of Generative AI: Threat to academic integrity? Journal of
Responsible Technology, 13, 1–4. https://doi.org/10.1016/j.jrt.2023.100060
Gill, S. S., & Kaur, R. (2023). ChatGPT: Vision and challenge. Internet of Things and Cyber-Physical
Systems, 3, 262–271. https://doi.org/10.1016/j.iotcps.2023.05.004
Hong, W. C. H. (2023). The impact of ChatGPT on foreign language teaching and learning: Opportunities
in education and research, Journal of Educational Technology and Innovation(JETI), 5(1),
37–45. https://jeti.thewsu.org/index.php/cieti/article/view/103/64
Kohnke, L., Moorhouse, B. L., & Zou, D. (2023). ChatGPT for language teaching and learning. RELC
Journal, 54(2), 1–14. https://doi.org/10.1177/00336882231162868
Meyer, J. G., et al. (2023). ChatGPT and large language models in academia: Opportunities and
challenges. BioData Mining 16(20). 1–11. https://doi.org/10.1186/s13040-023-00339-9
Miao, F. et al. (2021). Inteligencia artificial y educación: Guía para las personas a cargo de formular
políticas. UNESCO Publishing. https://unesdoc.unesco.org/ark:/48223/pf0000379376
Muñoz-Basols, J., & Fuertes Gutiérrez, M. (2024). Oportunidades de la Inteligencia Artificial (IA)
en el aprendizaje y la enseñanza de lenguas. In J. Muñoz-Basols, M. Fuertes Gutiérrez, & L.
Cerezo (Eds.), La enseñanza del español mediada por tecnología: de la justicia social a la
Inteligencia Artificial (IA), (ch.13). Routledge.
Peachey, N. (2023). ChatGPT in the language classroom. PeacheyPublications Ltd, 2023. https://
marcoele.com/chatgpt-como-agente-conversacional/
25 Technology in Language Teaching & Learning, 6(2)

Román-Mendoza, E. R. (2023). El arte de formular preguntas para comprender respuestas: ChatGPT


como agente conversacional en el aprendizaje de español como segunda lengua. marcoELE
Revista de didáctica ELE, 36, 1–18.
Rudolph, J., Tan, S., & Tan, S. (2023). ChatGPT: Bullshit spewer or the end of traditional assessments
in higher education? Journal of Applied Learning and Teaching, 6(1), 342–363. https://doi.
org/10.37074/jalt.2023.6.1.9
Sabzalieva, E., & Valentini, A. (2023). ChatGPT and artificial intelligence in higher education: Quick
start guide. UNESCO. https://www.iesalc.unesco.org/wp-content/uploads/2023/04/ChatGPT-
and-Artificial-Int elligence-in-higher-education-Quick-Start-guide_EN_FINAL.pdf
Stake, R. (2003). Case studies. In N. K. Denzin & Y. S. Lincoln (Eds.), Strategies of qualitative inquiry
(2nd ed., pp. 134–164). SAGE Publications.
Tenhundfeld, N., & ChatGPT (2023, February). Two Birds With One Stone: Writing a Paper Entitled
“ChatGPT as a Tool for Studying Human-AI Interaction in the Wild” with ChatGPT.
ResearchGate. https://doi.org/10.13140/RG.2.2.25319.73123
UNESCO (2019, May 16–18). Beijing Consensus on Artificial Intelligence and Education. [Paper
Presentation]. International Conference on Artificial Intelligence and Education, Planning
Education in the AI Era: Lead the Leap, Beijing, People’s Republic of China. https://unesdoc.
unesco.org/ark:/48223/pf0000368303
Zhu, C. et al. (2023). How to harness the potential of ChatGPT in education? Knowledge Management
& ELearning: An International Journal, 15(2), 133–152. https://doi.org/10.34105/j.
kmel.2023.15.008

You might also like